usp10: fix tests to allow atomic patches for new implementation
cjstimpson at utwire.net
Fri Dec 15 09:31:58 CST 2006
James Hawkins wrote:
> On 12/15/06, Clinton Stimpson <cjstimpson at utwire.net> wrote:
>> Here's a patch to make unimplemented functions return E_NOTIMPL instead
>> of S_OK. Even dependent functions already return E_NOTIMPL. Also
>> fixing tests to check that the return value is S_OK before testing the
>> 'out' parameters.
>> This fix allows piecemeal patches to follow with implementation for each
>> function in a separate patch.
>> Clinton Stimpson
>> Return E_NOTIMPL where appropriate.
>> And only check out parameters when appropriate.
> There are a couple things wrong with this patch. First, you've made
> it so the tests won't be run for these functions (because they don't
> return S_OK as it stands). Second, there are probably applications
> that depend on these functions returning S_OK. You're really making
> it a lot harder than it should be. Implement each function one at a
> time. If tests start to succeed, just remove the todo_wine from them.
> So the implemented function and removed/added go in one patch. You
> just do that for each function you're implementing.
Or, you guys are making it harder for me ;)
todo_wine ok( compare values from ScriptStringCPtoX with
So. My first patch implements ScriptStringAnalyse. Other ScriptString*
functions depend on that being implemented first.
So I go through and remove the todo_wine's where I get "test
succeeded". And the above todo_wine in the example above needs to be
removed. I do that, and then I get a bunch of "test failed" when i !=
0. It passes if i =0, but it fails with i != 0. The "make test"
returns a failure whether I leave or remove that todo_wine.
What am I supposed to do?
More information about the wine-devel